| Age | Commit message (Collapse) | Author |
|---|
|
As per the man page, the number of elements should be the first argument
and the size of each one the second.
Switch the arguments around.
The behavior will not change, as both arguments are only multiplied with
each other.
Reported-by: Sean Christopherson <seanjc@google.com>
Link: https://lore.kernel.org/lkml/aqQmCtPf_5M-BjRW@google.com/
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20260924-nolibc-calloc-v1-1-c49b383475d5@weissschuh.net
|
|
malloc() adds the size of its header to the requested size without
checking for overflow. For a size within the header size of SIZE_MAX,
the sum wraps around to a few bytes, and malloc() returns a single page
instead of NULL.
calloc() checks its multiplication for overflow, but then passes the
product to malloc(), so calloc(SIZE_MAX, 1) returns a single page too.
So does realloc(ptr, SIZE_MAX).
Such a size is usually a bug in the caller, for instance a length of -1
used as a size_t, and returning a page instead of NULL can turn it into
a heap overflow.
Fail with ENOMEM when adding the header overflows, as glibc does for
sizes this large.
Fixes: 0e0ff638400b ("tools/nolibc/stdlib: Implement `malloc()`, `calloc()`, `realloc()` and `free()`")
Assisted-by: LLM
Signed-off-by: Danish Khateeb <danishkhateeb03@gmail.com>
Link: https://patch.msgid.link/20260926152414.251214-3-danishkhateeb03@gmail.com
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
malloc() rounds the size of each mapping up to a multiple of 4096,
presumably to align it to a page boundary. But the page size is not 4096
on all architectures, and mmap() and munmap() round the length up to the
page size themselves.
Drop the rounding. As heap->len is no longer rounded up, growing a block
with realloc() now always creates a new mapping.
Assisted-by: LLM
Signed-off-by: Danish Khateeb <danishkhateeb03@gmail.com>
Reviewed-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260926152414.251214-2-danishkhateeb03@gmail.com
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
These are standard function to interact with a file descriptor without
touching its internal file offset.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Reviewed-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260927-nolibc-pread-v1-5-e9af6e47b37a@weissschuh.net
|
|
ILP32 architectures, like x32 and n32, support 64-bit system call
arguments, although they are nominally 32-bit architectures.
Add a new define which provides the size of system call arguments
so generic libc code can use the best system call for those.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Reviewed-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260927-nolibc-pread-v1-4-e9af6e47b37a@weissschuh.net
|
|
pread() and pwrite() support will require the usage of the macro from
the arch-*.h headers. sys.h can not be used from those.
Move the macro to a more generic header.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Reviewed-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260927-nolibc-pread-v1-3-e9af6e47b37a@weissschuh.net
|
|
Several 32-bit architectures pass a 64-bit systemcall argument
as two 32-bit values requiring alignment to an even register pair.
This case is currently relevant for the ftruncate64() wrapper and handled
by those architectures providing custom wrappers.
With the introduction of pread() and pwrite() wrappers, the same pattern
would lead to a lot of repeated code.
Add a #define, so that the generic code can do the padding where
necessary, without any per-architecture duplication.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Reviewed-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260927-nolibc-pread-v1-2-e9af6e47b37a@weissschuh.net
|
|
Extend the existing parisc support to also handle 64-bit mode.
Some peculiarities:
* The nonstandard ftruncate64 system call requires a workaround.
* The linker defaults to main() as entry point, which needs to be
switched to _start() explicitly.
* The linker defaults to dynamic linking, which needs to be disabled
explicitly.
* With the default -Os, gcc does not use 64-bit multiplication
instructions, requiring libgcc. Use -O2 instead.
* There is no qemu-user available.
* CONFIG_COMPAT needs to be disabled to avoid build errors due to a
missing 32-bit vDSO compiler.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20260914-nolibc-parisc64-v3-1-f14b320117c1@weissschuh.net
|
|
Since commit feaf75658783 ("nolibc: fix fd_set type"), an fd_set is an
array of unsigned long, but the FD_* macros still build their masks as
an int or an unsigned int: 1 << n in FD_SET(), 1U << n in FD_CLR() and
FD_ISSET(), where n is the fd modulo 64 on 64-bit architectures. So:
- FD_SET() of an fd with n == 31 sets the bits for n = 31-63, as
1 << 31 is negative and gets sign-extended.
- For n >= 32 the shifts are undefined. x86-64, for instance, masks
the shift count, so FD_SET(40) and FD_ISSET(40) use the bit of fd 8.
- FD_CLR() zero-extends ~(1U << n), so it also clears the bits for
n = 32-63, whichever fd it is given.
select() on fd 40, for instance, fails with EBADF, or watches fd 8
instead if that one is open.
Build the masks from 1UL.
Fixes: feaf75658783 ("nolibc: fix fd_set type")
Assisted-by: LLM
Signed-off-by: Danish Khateeb <danishkhateeb03@gmail.com>
Link: https://patch.msgid.link/20260925211600.119116-3-danishkhateeb03@gmail.com
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
readdir_r() stores the result of _sys_lseek() in an int. Directory
offsets are opaque cookies which can use all 64 bits: ext4, for
instance, gives 64-bit processes 63-bit hashes, and
0x7fffffffffffffff as the offset after the last entry. Truncated to an
int, such an offset is negative about half the time, and readdir_r()
takes it for an error.
Since commit 4ada5679f18d ("tools/nolibc/dirent: avoid errno in
readdir_r"), readdir_r() fails at the first entry whose offset has
bit 31 set, which on ext4 is usually one of the first few, and returns
the truncated offset as the error number. Before that, only -1 counted
as an error, which the last entry always hits: readdir_r() then
returned errno, usually 0, without filling in the entry, so the caller
got the previous entry a second time and never saw the last one.
32-bit processes get 31-bit hashes from ext4 and are not affected.
Keep the offset in an off_t.
Fixes: 665fa8dea90d ("tools/nolibc: add support for directory access")
Assisted-by: LLM
Signed-off-by: Danish Khateeb <danishkhateeb03@gmail.com>
Link: https://patch.msgid.link/20260925211600.119116-2-danishkhateeb03@gmail.com
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
verrx() passes its va_list to warnx() instead of vwarnx(). warnx() is
variadic, so it takes the va_list as its first and only argument, and
the conversions in the format string print the va_list itself and then
whatever is left in the following argument slots. On x86_64, for
instance,
errx(1, "int args: %d %d %d", 1, 2, 3);
prints something like "int args: -1699009032 1 2", where the first
number comes from the address of the va_list and changes from run to
run, and
errx(1, "%d %s", 1234, "foo");
crashes with SIGSEGV, as "%s" takes a leftover argument slot for the
string. errx() goes through verrx(), so every message with arguments
printed by either of them is wrong.
Call vwarnx() from verrx(), as verr() calls vwarn().
Fixes: 9da0f529c089 ("tools/nolibc: add err.h")
Assisted-by: LLM
Signed-off-by: Danish Khateeb <danishkhateeb03@gmail.com>
Link: https://patch.msgid.link/20260924120958.23698-2-danishkhateeb03@gmail.com
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
WIFSIGNALED() needs the subtraction to wrap around for status 0, so that
only 1 to 0xff, a terminating signal with or without the core dump flag,
pass the check. But status is an int, and 0 - 1 is -1, which is less
than 0xff. A child that exits with 0 is therefore reported as both
exited and killed, and a caller that tests WIFSIGNALED() first sees
"killed by signal 0".
Make the subtraction unsigned, as musl does in the same macro.
Fixes: 8c934d4822c7 ("tools/nolibc: add helpers for wait() signal exits")
Assisted-by: LLM
Signed-off-by: Danish Khateeb <danishkhateeb03@gmail.com>
Link: https://patch.msgid.link/20260923205622.1123270-2-danishkhateeb03@gmail.com
[Thomas: Drop stable@ Cc]
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
The parisc initialization code inconsistently uses both %dp and %r27 to
refer to the data pointer register.
Consistently use the %dp memnonic.
Suggested-by: Helge Deller <deller@gmx.de>
Link: https://lore.kernel.org/lkml/495ef25e-0bac-4f98-b0da-ae0d96dcea59@gmx.de/
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20260914-nolibc-parisc-dp-register-v1-1-6388f4aa8fbe@weissschuh.net
|
|
If a non-directory is opened, ENODIR should be returned from
opendir()/fdopendir() right away and not only during readdir_r().
Validate the type of opened file during open.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Reviewed-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260831-nolibc-fdopendir-enotdir-v1-1-8cf0e79c4e6f@weissschuh.net
|
|
Switch to O_DIRECTORY to validate that the opened file is a directory.
ALso remove the call to fdopendir() in opendir() to avoid a double check
of 'fd'. This will also make it easier to add a similar check for
fdopendir().
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Reviewed-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260831-nolibc-fdopendir-enotdir-v1-3-8cf0e79c4e6f@weissschuh.net
|
|
This is useful because it allows copying data up to 2GB from one
fd to another without a buffer in userspace and with a single syscall.
For 32bit kernels we use the sendfile64 syscall to allow for 64bit
offsets to work.
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Link: https://patch.msgid.link/20260828103205.384589-2-daniel@thingy.jp
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
A mostly straightforward new architecture with a few quirks:
* Only qemu-user is supported for testing.
* Clang is required for compilation.
* -fsanitize=undefined is broken.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Link: https://patch.msgid.link/20260819-nolibc-hexagon-v1-3-6bc3be591f09@weissschuh.net
|
|
The list is getting long, split it up for easier additions.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Reviewed-by: Brian Cain <brian.cain@oss.qualcomm.com>
Link: https://patch.msgid.link/20260819-nolibc-hexagon-v1-1-6bc3be591f09@weissschuh.net
|
|
A straightforward new architecture.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Tested-by: Michael Cree <mcree@orcon.net.nz>
Link: https://patch.msgid.link/20260722-nolibc-alpha-v2-1-4970e48eb7bf@weissschuh.net
|
|
The system call can clobber the ctr and xer registers.
Make sure the compiler takes this into account.
The missing clobbers only seem to be an issue with newer compilers.
Fixes: 0cb0675ec37e ("tools/nolibc: add support for powerpc")
Signed-off-by: Thomas Weißschuh <thomas.weissschuh@linutronix.de>
Link: https://patch.msgid.link/20260727-nolibc-powerpc-clobber-v1-1-e0911cc99ce1@linutronix.de
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
The macro lost its last user when select() was reimplemented on top of
pselect6 for every architecture; sys/select.h no longer looks at it,
and nothing else in the tree does either. See commit 668e43737279
("tools/nolibc/select: drop non-pselect based implementations").
The comment that goes with it is stale for the same reason, as neither
architecture can end up calling old_select any more.
Drop both from arch-x86.h and arch-arm.h.
Cc: Alviro Iskandar Setiawan <alviro.iskandar@gnuweeb.org>
Signed-off-by: Ammar Faizi <ammarfaizi2@openresty.com>
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
__nolibc_syscall0() does not set the arg1 variable before passing it to
the asm block. This uninitialized variable read is undefined behavior.
Clang can miscompile this.
Mark the asm operand as write-only to fix this.
Fixes: 8e1930296f92 ("tools/nolibc: Add support for SPARC")
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260703-nolibc-sparc-asm-v1-1-c7fe73e2e777@weissschuh.net
|
|
Add readlink(). This is needed to test getcwd().
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260702085101.3304547-3-daniel@thingy.jp
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
Add getcwd() for getting the current working directory.
The behaviour matches what musl is doing except for one
important difference: If the passed buf is NULL musl (and glibc)
uses a big buffer on the stack and that is then strdup()'d and
returned.
According to the man page for getcwd() this is a glibc extension
and I don't think we need it in nolibc.
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260702085101.3304547-2-daniel@thingy.jp
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
With -Wwrite-strings the plain assignment triggers a warning as a
'const char *' is assigned to a 'char *', removing the const qualifier.
Casting the const away is fine, as there is no valid modification that
can be done to an empty string anyways.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260525-nolibc-write-strings-v2-1-ab5cc16c7b23@weissschuh.net
|
|
On architectures with 32-bit longs, call the compat syscall
__NR_ftruncate64. As off_t is 64-bit it must be split into 2 registers.
Unlike llseek() which passes the high and low parts in explicitly named
arguments, the order here is endian independent.
Some architectures (arm, mips, ppc) require this pair of registers to
be aligned to an even register, so add custom _sys_ftruncate64()
wrappers for those.
A test case for ftruncate is added which validates negative length or
invalid fd return the appropriate error, and checks the length is
correct on success.
Co-developed-by: Jordan Richards <jordanrichards@google.com>
Signed-off-by: Jordan Richards <jordanrichards@google.com>
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Reviewed-by: Daniel Palmer <daniel@thingy.jp>
Link: https://patch.msgid.link/20260521-nolibc-ftruncate-v1-3-5384a83b2402@weissschuh.net
|
|
On 32-bit architectures some system calls require a single 64-bit
argument to be passed as two 32-bit halves.
Add a helper to easily split such arguments. This works on little and
bit endian.
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260521-nolibc-ftruncate-v1-2-5384a83b2402@weissschuh.net
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
init yet
We are using the getrandom syscall to get a random seed for the
stack protector canary but we are calling it with no flags which means
it'll block until there is some real randomness to return.
This means that if the crng is not ready yet program startup will
block and if you are unlucky that could be for a long time and
look like the program has crashed.
Even if the call to getrandom does not yield any random data,
we will still initialize the canary.
Fixes: 7188d4637e95 ("tools/nolibc: add support for stack protector")
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260522090726.726985-1-daniel@thingy.jp
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
Running clang-tidy on a program that uses getopt() from nolibc
this warning appears:
getopt.h:80:6: warning: Out of bound access to memory after the end of the string literal [clang-analyzer-security.ArrayBound]
80 | if (optstring[i] == ':') {
This looks like a very unlikely case that an argument
inside of argv is being changed between getopt() calls.
Adding a check for d becoming 0 in the guard after the loop
stops getopt() getting far enough to access beyond the end
of the array and seems to correct the issue.
Fixes: bae3cd708e8a ("tools/nolibc: add getopt()")
Assisted-by: Claude:claude-4.6-sonnet # reproducer
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Link: https://patch.msgid.link/20260520111931.1027758-1-daniel@thingy.jp
[Thomas: clean up commit message a bit]
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
When O_TMPFILE is set, the open mode needs to be passed to the kernel as
per the documentation. Currently this is not done.
Instead of checking for O_TMPFILE explicitly and making the conditionals
more complex, just always pass the mode to the kernel. If no value was
passed the mode will be garbage, but the kernel will ignore it anyways.
Fixes: a7604ba149e7 ("tools/nolibc/sys: make open() take a vararg on the 3rd argument")
Suggested-by: Willy Tarreau <w@1wt.eu>
Link: https://lore.kernel.org/lkml/afRfjdovT6pNtwtP@1wt.eu/
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260514-nolibc-open-tmpfile-v2-3-b4c6c5efa266@weissschuh.net
|
|
This logic is duplicated and some upcoming extensions would require even
more duplicated logic.
Move it into a macro to avoid the duplication and allow cleaner changes.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260514-nolibc-open-tmpfile-v2-2-b4c6c5efa266@weissschuh.net
|
|
This logic is duplicated and its current form will be in the way of some
upcoming simplificiations.
Move it into a macro to avoid the duplication and enable some cleanups.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260514-nolibc-open-tmpfile-v2-1-b4c6c5efa266@weissschuh.net
|
|
Extend nolibc to target the 32-bit parisc architecture.
64-bit is not yet supported.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch/msgid.link/20260428-nolibc-hppa-v5-2-d843d573111a@weissschuh.net
|
|
Add support for OpenRISC / or1k to nolibc.
_start() uses the same wrapper construct as in arch-sh.h.
libgcc is necessary as OpenRISC is missing 64-bit multiplication.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Stafford Horne <shorne@gmail.com>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260429-nolibc-openrisc-v2-1-8d7d7a2f3fec@weissschuh.net
|
|
nolibc can natively handle large files. Tell this to the kernel by
always using O_LARGEFILE when opening files. This is also how other
libcs do it.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260418-nolibc-largefile-v1-6-b91f0775bac3@weissschuh.net
|
|
The N32 system call ABI expects 64-bit values directly in registers.
This does not work on nolibc currently, as a 'long' is only 32 bits
wide. Switch the system call wrappers to use 'long long' instead which
can handle 64-bit values on N32. As on N64 'long' and 'long long' are
the same, this does not change the behavior there.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260418-nolibc-largefile-v1-5-b91f0775bac3@weissschuh.net
|
|
The x32 system call ABI expects 64-bit values directly in registers.
This does not work on nolibc currently, as a 'long' is only 32 bits
wide. Switch the system call wrappers to use 'long long' instead which
can handle 64-bit values on x32. As on x86_64 'long' and 'long long' are
the same, this does not change the behavior there.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260418-nolibc-largefile-v1-4-b91f0775bac3@weissschuh.net
|
|
Currently all system call wrappers return 'long' integers which can be
directly cast to 'void *' if the returned value is actually a pointer.
An upcoming change will change the system call wrappers to sometimes
return 'long long' which can not be cast to a pointer directly.
Add explicit cast through 'long' to prepare for this.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260418-nolibc-largefile-v1-3-b91f0775bac3@weissschuh.net
|
|
In the architecture specific system call glue, all arguments are
currently casted to 'long' to fit into registers. This works for
pointers as 'long' has the same size as pointers.
However the system call registers for X32 and MIPS N32 need to be
'long long' to work correctly for 64-bit values expected by the system
call ABI. Casting a pointer to a 'long long' will produce a compiler
warning while casting 64-bit integers to 'long' will truncate those.
Add a helper which can be used to correctly cast both pointers and
integers into 'long long' registers. Cast the pointers through
'unsigned' to avoid any sign extensions.
Both builtins have been available since at least GCC 3 and clang 3.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260418-nolibc-largefile-v1-2-b91f0775bac3@weissschuh.net
|
|
On some architectures the llseek system call contains a leading
underscore. Treat it the same way as llseek and prefer it over the
plain lseek system call as is necessary for 64-bit offset handling.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260418-nolibc-largefile-v1-1-b91f0775bac3@weissschuh.net
|
|
creat() is a simple wrapper around open().
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260419-nolibc-open-mode-v1-2-8dc5a960daa7@weissschuh.net
|
|
The stack protector is configured in _start_c() so we shouldn't
use it before then.
Add __nolibc_no_stack_protector to _start_c() to avoid the compiler
generating stack protector code for _start_c() and thus using it
before its configured.
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Link: https://patch.msgid.link/20260425111315.3191461-3-daniel@thingy.jp
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
To avoid polluting the namespace rename __no_stack_protector to
__nolibc_no_stack_protector so its now within the nolibc umbrella.
Suggested-by: Willy Tarreau <w@1wt.eu>
Signed-off-by: Daniel Palmer <daniel@thingy.jp>
Link: https://patch.msgid.link/20260425111315.3191461-2-daniel@thingy.jp
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
|
|
Clang may convert the loop to find _auxv into a call to wcslen() which
is missing on nolibc. -fsanitize needs to be disabled for this to
happen.
Use the same pattern as in the nolibc strlen() implementation to avoid
the function call generation.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260418-nolibc-wcslen-v1-1-671271b8ea63@weissschuh.net
|
|
Functions which are known at compile-time to result in ENOSYS can be
surprising to the user. For example using old UAPI headers might mean
that stat() will always fail although the kernel would have the system
call available at runtime. Nowadays __nolibc_enosys() should never be
called for normal applications.
Switch the silent ENOSYS return into a compile-time error, so the user
is aware about the issue. Prefer the 'error' attribute as it provides
the best diagnostics. If the users defines NOLIBC_COMPILE_TIME_ENOSYS
the old, silent fallback is kept.
Also add a test which validates that the error can be optimized away.
Reported-by: Willy Tarreau <w@1wt.eu>
Closes: https://lore.kernel.org/lkml/acizRIq2xrFUNHNS@1wt.eu/
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260404-nolibc-enosys-v1-1-e0aba47bdee4@weissschuh.net
|
|
Add the wide-used alloca() function. As it is highly machine and
compiler dependent, just defer to the compiler builtin. This has
been available since GCC 4 and clang 3.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260409-nolibc-alloca-v1-1-ed02f68dfaf9@weissschuh.net
|
|
Add the standard assert() macro from the assert.h header.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260409-nolibc-assert-v1-1-42da8b367e23@weissschuh.net
|
|
The UBSAN implementation in gcc requires a slightly different function
attribute to skip instrumentation.
Extend __nolibc_no_sanitize_undefined to also handle gcc.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20260408-nolibc-gcc-15-v1-2-330d0c40f894@weissschuh.net
|
|
The logic to disable UBSAN will become a bit more complicated.
Move it out into compiler.h, so crt.h stays readable.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Link: https://patch.msgid.link/20260408-nolibc-gcc-15-v1-1-330d0c40f894@weissschuh.net
|
|
Add some standard functions to convert between different byte orders.
Conveniently the UAPI headers provide all the necessary functionality.
Signed-off-by: Thomas Weißschuh <linux@weissschuh.net>
Acked-by: Willy Tarreau <w@1wt.eu>
Link: https://patch.msgid.link/20260405-nolibc-bswap-v1-1-f7699ca9cee0@weissschuh.net
|
